I have an LSTM model in python and I am trying to deploy it on ZCU104 board using Vitis AI(Pytorch). I am stuck on an error while quantizing the model. I am getting the error for the line:

quantizer.export_xmodel(output_dir=”quantize_result”, deploy_check=True)

and the error is:

[VAIQ_NOTE]: =>Converting to xmodel ...

Traceback (most recent call last):

  File "lstm_quant.py", line 118, in <module>

    main(args)

  File "lstm_quant.py", line 107, in main

    quantizer.export_xmodel(output_dir="quantize_result", deploy_check=True)

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/pytorch_nndct/apis.py", line 148, in export_xmodel

    self.processor.export_xmodel(output_dir, deploy_check, dynamic_batch)

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/pytorch_nndct/qproc/base.py", line 368, in export_xmodel

    dump_xmodel(output_dir, deploy_check, self._lstm_app)

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/pytorch_nndct/qproc/base.py", line 505, in dump_xmodel

    deploy_graphs, _ = get_deploy_graph_list(quantizer.quant_model, quantizer.Nndctgraph)

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/pytorch_nndct/qproc/utils.py", line 463, in get_deploy_graph_list

    return _deploy_optimize(quant_model, nndct_graph, need_partition)

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/pytorch_nndct/qproc/utils.py", line 419, in _deploy_optimize

    g_optmizer = DevGraphOptimizer(nndct_graph)

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/nndct_shared/compile/deploy_optimizer.py", line 92, in __init__

    self._dev_graph.clone_from(nndct_graph)

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/nndct_shared/nndct_graph/base_graph.py", line 133, in clone_from

    self._top_block.clone_from(src_graph.block, local_map, converted_nodes)

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/nndct_shared/nndct_graph/base_block.py", line 47, in clone_from

    self.append_node(self.owning_graph.create_node_from(node, local_map, converted_nodes))

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/nndct_shared/nndct_graph/base_graph.py", line 161, in create_node_from

    node.clone_from(src_node, local_map)

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/nndct_shared/nndct_graph/base_node.py", line 120, in clone_from

    self.op.clone_from(src_node.op, local_map)

  File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.8/site-packages/nndct_shared/nndct_graph/base_operator.py", line 214, in clone_from

    setattr(self, config, new_value)

AttributeError: can't set attribute

Attaching the full error, model code and quantization script in the google links below:

Quantization Script:
https://docs.google.com/document/d/1jRYmPH2z70ovpc_FJIBpUQaTHUPRrTgnPVxlQJ1JLug/edit?usp=sharing

Model Script:
https://docs.google.com/document/d/1OBZw4XhdHpVhA0gKcJn2NzR_WA7ZczQ3hL42f_sMKug/edit?usp=sharing

Full Error:
https://docs.google.com/document/d/1kI1WJqq9pp3aSsGLpNGjf22mzK6swIiZbIXwfUeTGto/edit?usp=sharing

Tried changing the base_operator.py but didn’t work. Also tried simplifing the model but didn’t work.